Goodness! This is amazing 😀🙏 It's going to take me a minute to digest this all and find a way forward in terms of optimised port placement and geometry but its really interesting to see the different simulations!
Fair enough - I was being lazy and not taking all the cad screen shots. I can't see a way to edit my last past, so here's the entire post with cad added. I had to split it across multiple posts as there are too many images. Sorry if the formatting gets messed up - I'm pasting this in.
Here are some results from 3D FEA simulations of some aspects of midrange geometry. I started off modeling rrobot's geometry from hornresp as a square horn with his port geometry and a rough front chamber based on the available dimensions of the earthquake driver. Then I changed a bunch of things to see how the response would be effected. I'm modeling only the air volumes in and around the speaker and applying a constant acceleration uniformly across the driver's cone. There's no attempt to model any effects of the driver's response due to cone breakup or its electro-mechanical parameters. What this type of data is good for is seeing the effects of the 3 dimensional geometry you're putting the driver in and seeing relative changes between different geometries. It's also good for looking at directivity, but I didn't pull that data out here. I'm just showing the response in free space at about 0.5m in front of the mouth of the horn. Here's what one of the geometries looks like plus a closer view of the air in front of the driver and the port. The highlighted blue areas are where I've applied the acceleration to the cone (along the driver's axis) as a stimulus for the simulation. It's a 1/4 symmetry model, so even though there's only 1 driver the model is simulated 4 drivers symmetrically arranged around the horn.
First I looked at a rectangular port entering the horn tangent to one of the walls and compared that to an equal area round port. Here's a plot showing the simulated responses on top and the difference between them on the bottom. All these plots run from 200Hz to 4kHz. So it looks like a rectangular port has about 0.5dB more output than a round port and shifts around the resonances in the horn slightly.
Rectangular port (all the cad screen shots will show one driver for better visibility, but the simulations are all with four drivers). The port isn’t technically rectangular - two sides are parallel to the wall of the horn and the other two sides (front and back) are parallel to the horn mouth.
Round port:
Next I looked at the rectangular port (more or less square - 20x23mm) vs an equal area but long and narrow port (so about twice the length along the corner of the horn but half as wide). The narrow port has about 0.5dB more output at higher frequencies (by which I mean around 600-1.5kHz).
Narrow port:
Here's the original rectangular port which comes in perpendicular to the wall of the horn compared to the same port but coming in at a 45 degree angle to the horn wall (it's actually only 45 degrees when viewed from the front or back). This is what's shown in the example model image. This seems to pickup 1.5-2dB which is interesting. The length is the same going from the point of the corner, so arguably the port length is changing slightly.
Port coming in at 45 degrees:
The ports entering perpendicular to the horn wall force the ports to be somewhat close to the perimeter of the driver in order to not have the drivers run into each other. The port entering at 45 degrees doesn't do that, so the port ended up closer to the driver's center. Moving the driver so the port enters the front chamber closer to the edge of the driver gives about a 3dB+ reduction in output.
Driver moved to have port closer to the driver edge:
If I take that 45 degree port and flare its entrance towards the driver's center, that gives me an extra 1dB of output.
Flare port entrance:
The 45 degree entrance port generally seems like a winner so let's keep going with that. If I put a filler plug in the front chamber to follow the surface of the cone, that makes the port longer (the driver stays in the same location, so now the port has to extend further to reach the front chamber). But having the small front chamber shifts the resonances around in the system quite a bit and moves the first pressure null in the front chamber down from around 2k to 1.5kHz. You also get a lot more higher frequency output (2-4kHz) which I would consider a negative given you wouldn't want to use this above maybe 1.3kHz.
Driver with front chamber filled in. I cut through this cad at a different angle to show how the port intersects the front chamber and shape of the front chamber.
If I take that smaller front chamber geometry and flare the port inlet like I've seen people do, that shifts that null up a bit but doesn't seem to help a ton.
Small front chamber with the inlet port flared (but only in one plane). Cad is cut through the center of the driver here.
Here are some results from 3D FEA simulations of some aspects of midrange geometry. I started off modeling rrobot's geometry from hornresp as a square horn with his port geometry and a rough front chamber based on the available dimensions of the earthquake driver. Then I changed a bunch of things to see how the response would be effected. I'm modeling only the air volumes in and around the speaker and applying a constant acceleration uniformly across the driver's cone. There's no attempt to model any effects of the driver's response due to cone breakup or its electro-mechanical parameters. What this type of data is good for is seeing the effects of the 3 dimensional geometry you're putting the driver in and seeing relative changes between different geometries. It's also good for looking at directivity, but I didn't pull that data out here. I'm just showing the response in free space at about 0.5m in front of the mouth of the horn. Here's what one of the geometries looks like plus a closer view of the air in front of the driver and the port. The highlighted blue areas are where I've applied the acceleration to the cone (along the driver's axis) as a stimulus for the simulation. It's a 1/4 symmetry model, so even though there's only 1 driver the model is simulated 4 drivers symmetrically arranged around the horn.
First I looked at a rectangular port entering the horn tangent to one of the walls and compared that to an equal area round port. Here's a plot showing the simulated responses on top and the difference between them on the bottom. All these plots run from 200Hz to 4kHz. So it looks like a rectangular port has about 0.5dB more output than a round port and shifts around the resonances in the horn slightly.
Rectangular port (all the cad screen shots will show one driver for better visibility, but the simulations are all with four drivers). The port isn’t technically rectangular - two sides are parallel to the wall of the horn and the other two sides (front and back) are parallel to the horn mouth.
Round port:
Next I looked at the rectangular port (more or less square - 20x23mm) vs an equal area but long and narrow port (so about twice the length along the corner of the horn but half as wide). The narrow port has about 0.5dB more output at higher frequencies (by which I mean around 600-1.5kHz).
Narrow port:
Here's the original rectangular port which comes in perpendicular to the wall of the horn compared to the same port but coming in at a 45 degree angle to the horn wall (it's actually only 45 degrees when viewed from the front or back). This is what's shown in the example model image. This seems to pickup 1.5-2dB which is interesting. The length is the same going from the point of the corner, so arguably the port length is changing slightly.
Port coming in at 45 degrees:
The ports entering perpendicular to the horn wall force the ports to be somewhat close to the perimeter of the driver in order to not have the drivers run into each other. The port entering at 45 degrees doesn't do that, so the port ended up closer to the driver's center. Moving the driver so the port enters the front chamber closer to the edge of the driver gives about a 3dB+ reduction in output.
Driver moved to have port closer to the driver edge:
If I take that 45 degree port and flare its entrance towards the driver's center, that gives me an extra 1dB of output.
Flare port entrance:
The 45 degree entrance port generally seems like a winner so let's keep going with that. If I put a filler plug in the front chamber to follow the surface of the cone, that makes the port longer (the driver stays in the same location, so now the port has to extend further to reach the front chamber). But having the small front chamber shifts the resonances around in the system quite a bit and moves the first pressure null in the front chamber down from around 2k to 1.5kHz. You also get a lot more higher frequency output (2-4kHz) which I would consider a negative given you wouldn't want to use this above maybe 1.3kHz.
Driver with front chamber filled in. I cut through this cad at a different angle to show how the port intersects the front chamber and shape of the front chamber.
If I take that smaller front chamber geometry and flare the port inlet like I've seen people do, that shifts that null up a bit but doesn't seem to help a ton.
Small front chamber with the inlet port flared (but only in one plane). Cad is cut through the center of the driver here.
However if I take that small front chamber and make a single slot circumferential phase plug going into the rectangular port, I can smooth out that null from the front chamber. I just drew something here - no calculations, so I'd assume I could smooth this out more if I actually spent some time on it.
Single slot phase plug port cut through the center of the driver:
If I take the original 45 degree port and change it to be the longer narrower port, I get a marginal improvement of <0.5dB. Just speculating but maybe the improvement from round to rectangular to narrow to 45 degree entrance is all on a spectrum of getting improvements from getting the port into the corner and here we've gotten the last of that.
Longer narrower port entering on the corner of the horn:
Here's looking at different inlet port flares on the 45 degree long narrow port. First I flared it in only the plane along the corner of the horn towards the driver center, good for about 0.5dB. Then I additionally flared it to both sides so opening across more of the driver cone which gave a 2-3dB improvement at high frequencies. Then I made a lofted surface that flared the port inlet out to a circle equal to the diameter of the surround of the driver. This gave more high frequency gain but less gain below 1kHz and lower compared to the port flared horizontally, so depending on what you're going for maybe there's a happy medium between those two that could be optimized for.
Flaring in one plane:
Flaring towards the front and both sides:
Flaring to expand out to the diameter of the surround:
Then I took that geometry where the port flares to cover the entire driver cone and tried different port lengths (roughly half and double the port length I had been simulating). This shifts the front chamber resonance around 2kHz and the low frequency resonance around which also effects the output across most of the pass band. I would say shorter seems better. I think this is more changing the front chamber volume with this type of flare.
Half port length:
Double port length:
Last, I tried varying the length of the section of horn between the midrange ports and the small end of the horn. This one surprised me. I expected to see more of an effect of nulls in the response moving around. You do see that just below 2kHz but not a very big shift. For the half length geometry, I cut off the small end of the horn midway between the midrange port entrance and the small end of the horn. For the longer length, I actually drew in a completely made up 2 slot phase plug going out to a 50mm tweeter diaphragm and front chamber. Note that my frequency spacing is not super tight so I wouldn't read too much into the notches in the response being deeper or shallower around 2kHz. It could just be an effect of where the nulls in the response fell in relation to where my simulated frequency points were.
Half of the horn cut off between the midrange port entrance and the tweeter:
Tweeter phase plug added on:
Anyway, I hope someone finds this useful. I certainly found it interesting. I might have to hack the back off my old unity horn and 3d print something to try out some of these ideas. Also, this is only half to one third of the full picture. The other interesting thing to look at would be the effects of all of this on the tweeter's response. I need to finish building a more powerful computer to be able to look at that. It could also be interesting to look at air velocity levels through these different ports (which would give different max output limits) although they're all the same area so I would imagine there wouldn't be huge differences there.
Single slot phase plug port cut through the center of the driver:
If I take the original 45 degree port and change it to be the longer narrower port, I get a marginal improvement of <0.5dB. Just speculating but maybe the improvement from round to rectangular to narrow to 45 degree entrance is all on a spectrum of getting improvements from getting the port into the corner and here we've gotten the last of that.
Longer narrower port entering on the corner of the horn:
Here's looking at different inlet port flares on the 45 degree long narrow port. First I flared it in only the plane along the corner of the horn towards the driver center, good for about 0.5dB. Then I additionally flared it to both sides so opening across more of the driver cone which gave a 2-3dB improvement at high frequencies. Then I made a lofted surface that flared the port inlet out to a circle equal to the diameter of the surround of the driver. This gave more high frequency gain but less gain below 1kHz and lower compared to the port flared horizontally, so depending on what you're going for maybe there's a happy medium between those two that could be optimized for.
Flaring in one plane:
Flaring towards the front and both sides:
Flaring to expand out to the diameter of the surround:
Then I took that geometry where the port flares to cover the entire driver cone and tried different port lengths (roughly half and double the port length I had been simulating). This shifts the front chamber resonance around 2kHz and the low frequency resonance around which also effects the output across most of the pass band. I would say shorter seems better. I think this is more changing the front chamber volume with this type of flare.
Half port length:
Double port length:
Last, I tried varying the length of the section of horn between the midrange ports and the small end of the horn. This one surprised me. I expected to see more of an effect of nulls in the response moving around. You do see that just below 2kHz but not a very big shift. For the half length geometry, I cut off the small end of the horn midway between the midrange port entrance and the small end of the horn. For the longer length, I actually drew in a completely made up 2 slot phase plug going out to a 50mm tweeter diaphragm and front chamber. Note that my frequency spacing is not super tight so I wouldn't read too much into the notches in the response being deeper or shallower around 2kHz. It could just be an effect of where the nulls in the response fell in relation to where my simulated frequency points were.
Half of the horn cut off between the midrange port entrance and the tweeter:
Tweeter phase plug added on:
Anyway, I hope someone finds this useful. I certainly found it interesting. I might have to hack the back off my old unity horn and 3d print something to try out some of these ideas. Also, this is only half to one third of the full picture. The other interesting thing to look at would be the effects of all of this on the tweeter's response. I need to finish building a more powerful computer to be able to look at that. It could also be interesting to look at air velocity levels through these different ports (which would give different max output limits) although they're all the same area so I would imagine there wouldn't be huge differences there.
One more - if you compare a simple round port to the highest output of the non-phase plug ports, that's the 45 degree entrance, long narrow cross section port with a short length flaring to the full diameter of the surround. The difference is about 5dB more output from ~500Hz to 2kHz for the same cone acceleration. You do give up some output on the low end though so it depends on where you are challenged for output as to what might be more appealing.
we've all been tucking our small mid drivers behind the CD thinking we needed to minimize port length. But if an extra cm can be tolerated, then I think the drawing below shows a way to implement such that the port runs from under the center of the driver to the best location on the horn. With this, you can have 4NDF34's on all 4 sides with a horn mouth of 232 mm square or so. Port length would be just over 1". CD mounting plate sized for BMS5531.
This has been on my todo list for a while. Maybe I'll finally get off the dime and at least ABEC it
This has been on my todo list for a while. Maybe I'll finally get off the dime and at least ABEC it
I am currently playing with Visaton 5 FRSXs for use in a MEH and have produced acceptable results frequencyrange wise. I simulated a simple bandpass in WinISD and used a port resonance at 2 kHz. Rear volume or treatment doesn´t make to much difference in my case in testing at least.
I experience stupid high distortion levels though. Any recommendation for dealing with that? Can you treat the backside or front volume in any way that makes sense?
I want to use it from 400 Hz upwards at least, but this is measured from 20cm max and level isn´t that high. Measurements taken in a cardboard box, i have no idea how much that contributes to distortion. Backside is tightly wrapped in polyesterwool.
I
I experience stupid high distortion levels though. Any recommendation for dealing with that? Can you treat the backside or front volume in any way that makes sense?
I want to use it from 400 Hz upwards at least, but this is measured from 20cm max and level isn´t that high. Measurements taken in a cardboard box, i have no idea how much that contributes to distortion. Backside is tightly wrapped in polyesterwool.
I
Things are getting stranger... I have printed a full version of my waveguide and now have significant distortion levels at basically no exzessive levels. Looks like there is some form of crazy cone beakup going on. Anyone got some ideas to troubleshoot this?
Its basically the same Bandpassgeometry as before.
This was the rough bandpass test:
This is the backside of the waveguide. I have polyesterwool wrapped around the drivers, but it doesn´t make any difference.
Its basically the same Bandpassgeometry as before.
This was the rough bandpass test:
This is the backside of the waveguide. I have polyesterwool wrapped around the drivers, but it doesn´t make any difference.
Divide and conquer. Divide up the problem so you can isolate the cause. I would take the driver out of the bandpass and test it by itself to start. If the distortion disappears, it's something about the enclosure - design or construction. I'd also test at multiple drive levels to see if the distortion is level dependent (typically it would be).
It´s quite simply the port resonance (3kHz in my case).
Is putting the port resonance into the notch between throat and ports worth the hustle? Or do you just put the resonance so far beyond the passband that K2 is filtered out electrically? The only ways i can imagine making that work. If i want to sell my NS1000s, clean distortion mids is something i need to achieve 😛
Is putting the port resonance into the notch between throat and ports worth the hustle? Or do you just put the resonance so far beyond the passband that K2 is filtered out electrically? The only ways i can imagine making that work. If i want to sell my NS1000s, clean distortion mids is something i need to achieve 😛